Discussion:
[lfs-support] Boot freezes
Frans de Boer
2018-09-27 19:21:02 UTC
Permalink
LS,

I have a problem which I don't seem to be able to solve without further
help.

Using the development version for LFS, I get stuck every time during
boot time. It happens either after the message
"Fast init done" or right thereafter - with a delay of aprox 12 seconds
- at the message "Fast crng done".

This problem is with the standard LFS as well as the LFS-systemd.

Anyone has a suggestion?

Rgards, Frans.
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

http://en.wikipedia
Ken Moffat
2018-09-27 19:37:35 UTC
Permalink
Post by Frans de Boer
LS,
I have a problem which I don't seem to be able to solve without further
help.
Using the development version for LFS, I get stuck every time during boot
time. It happens either after the message
"Fast init done" or right thereafter - with a delay of aprox 12 seconds - at
the message "Fast crng done".
This problem is with the standard LFS as well as the LFS-systemd.
Anyone has a suggestion?
Rgards, Frans.
Off the top of my head, I'll guess lack of entropy. Changes were
made in the previous 2 or 3 kernel versions to fix vulnerabilities,
particularly related to generating secure secrets, but with the side
effect that booting can take longer (*much* longer in VMs).

The reals question is "what starts after the delay ?" For the
entropy problem, it would need to be something using random seeds.
But 12 seconds is not a long time for the entropy problem - are you
thumping the keyboards during that time ? (that adds to the
entropy).

So, maybe it is something else such as dhcp. Without knowing what
starts next, this is all guess work. Following comments are only
relevant if a lack of entropy is indeed the problem.

On my desktop machines, those with only SSD(s) and integrated
graphics had the entropy problem, those with spinning rust - and the
one with an SSD and a separate video card - did not.

To generate entropy, haveged is the current answer. With 4.19
kernels there should be an option to trust the hardware RNG (if you
have one).

ĸen
--
Tout est bien, tout va bien, tout va pour le mieux qu'il soit possible
-- Candide, de Voltaire
(Everything is for the best, in the best of all possible worlds)
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-ma
Bruce Dubbs
2018-09-27 20:01:49 UTC
Permalink
Post by Frans de Boer
LS,
I have a problem which I don't seem to be able to solve without further
help.
Using the development version for LFS, I get stuck every time during
boot time. It happens either after the message
"Fast init done" or right thereafter - with a delay of aprox 12 seconds
- at the message "Fast crng done".
This problem is with the standard LFS as well as the LFS-systemd.
Anyone has a suggestion?
It sounds like a kernel configuration issue. For your message is seems
that it doesn't even get to /sbin/init which is where the userspace is
initiated.

You don't say what version of the kernel you are using, but I suppose it
is it 4.18.9. Did you do 'make modules_install' ? You might want to
try an earlier known good kernel version.

-- Bruce
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

http:
Frans de Boer
2018-09-27 20:37:09 UTC
Permalink
Post by Frans de Boer
LS,
I have a problem which I don't seem to be able to solve without
further help.
Using the development version for LFS, I get stuck every time during
boot time. It happens either after the message
"Fast init done" or right thereafter - with a delay of aprox 12
seconds - at the message "Fast crng done".
This problem is with the standard LFS as well as the LFS-systemd.
Anyone has a suggestion?
It sounds like a kernel configuration issue.  For your message is
seems that it doesn't even get to /sbin/init which is where the
userspace is initiated.
You don't say what version of the kernel you are using, but I suppose
it is it 4.18.9.  Did you do 'make modules_install' ?  You might want
to try an earlier known good kernel version.
  -- Bruce
Ken and Bruce,

I suspected it has to do something with entropy, so I used 4.9.88 -
4.9.2 dit not compiled - and the result was the same. So, since 4.9.88
is from somewhere in March I assume that there is no provision yet for
the spectre etc. issue.

And yes, of course I did "modules_install", the modules are installed
and the initramfs is generated properly too. Until August all worked
well, but after August it did not anymore.

As far as I understand /sbin/init is called after enough random data has
been collected. I just remember that sysvinit has been upgraded too to
2.90 instead of 2.89 with patches. Might a problem there? I will
certainly give it a try to use 2.89 again, just to satisfy my curiosity ;)

I also will make myself familiar with debugging the boot procedure.
Somewhere on the Internet there must be a good tutorial to find.

--- Frans.
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

http://en.wikipedia.org/wiki
Ken Moffat
2018-09-27 22:07:24 UTC
Permalink
Post by Frans de Boer
Ken and Bruce,
I suspected it has to do something with entropy, so I used 4.9.88 - 4.9.2
dit not compiled - and the result was the same. So, since 4.9.88 is from
somewhere in March I assume that there is no provision yet for the spectre
etc. issue.
4.9 ? Really ? Rather you than me, I will be surprised if it works
with gcc-8.2.
Post by Frans de Boer
And yes, of course I did "modules_install", the modules are installed and
the initramfs is generated properly too. Until August all worked well, but
after August it did not anymore.
We seem to be at cross purposes. You said you were using the
development book, now you mention an initramfs so clearly you have
at least something from BLFS. So, specifying the version(s) of the
books might be relevant. I've only booted (sysv) later than 8.3 on
two systems, one with 4.14.71, the other with 4.18.9, using
LFS-20180920 and BLFS-20180915 with occasional newer versions,
and neither have the entropy problem.

BUT: with 4.18.x and your delay of 12 seconds - What boot messages
(sysv) follow after that 12 second delay (and yes, just a summary of
WHICH bootscript runs after the delay - I know what it is like trying
to write down the messages once the bootscripts are running.)

ĸen
--
Tout est bien, tout va bien, tout va pour le mieux qu'il soit possible
-- Candide, de Voltaire
(Everything is for the best, in the best of all possible worlds)
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

http://en.wikipedia.org/wiki/Posting_styl
Bruce Dubbs
2018-09-28 04:48:25 UTC
Permalink
Post by Ken Moffat
Post by Frans de Boer
Ken and Bruce,
I suspected it has to do something with entropy, so I used 4.9.88 - 4.9.2
dit not compiled - and the result was the same. So, since 4.9.88 is from
somewhere in March I assume that there is no provision yet for the spectre
etc. issue.
4.9 ? Really ? Rather you than me, I will be surprised if it works
with gcc-8.2.
Post by Frans de Boer
And yes, of course I did "modules_install", the modules are installed and
the initramfs is generated properly too. Until August all worked well, but
after August it did not anymore.
We seem to be at cross purposes. You said you were using the
development book, now you mention an initramfs so clearly you have
at least something from BLFS. So, specifying the version(s) of the
books might be relevant. I've only booted (sysv) later than 8.3 on
two systems, one with 4.14.71, the other with 4.18.9, using
LFS-20180920 and BLFS-20180915 with occasional newer versions,
and neither have the entropy problem.
BUT: with 4.18.x and your delay of 12 seconds - What boot messages
(sysv) follow after that 12 second delay (and yes, just a summary of
WHICH bootscript runs after the delay - I know what it is like trying
to write down the messages once the bootscripts are running.)
cat /var/log/boot.log

-- Bruce
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
Pierre Labastie
2018-09-28 07:16:20 UTC
Permalink
Post by Bruce Dubbs
Post by Frans de Boer
Ken and Bruce,
I suspected it has to do something with entropy, so I used 4.9.88 - 4.9.2
dit not compiled - and the result was the same. So, since 4.9.88 is from
somewhere in March I assume that there is no provision yet for the spectre
etc. issue.
4.9 ?  Really ?  Rather you than me, I will be surprised if it works
with gcc-8.2.
Post by Frans de Boer
And yes, of course I did "modules_install", the modules are installed and
the initramfs is generated properly too. Until August all worked well, but
after August it did not anymore.
We seem to be at cross purposes.  You said you were using the
development book, now you mention an initramfs so clearly you have
at least something from BLFS.  So, specifying the version(s) of the
books might be relevant.  I've only booted (sysv) later than 8.3 on
two systems, one with 4.14.71, the other with 4.18.9, using
LFS-20180920 and BLFS-20180915 with occasional newer versions,
and neither have the entropy problem.
BUT: with 4.18.x and your delay of 12 seconds - What boot messages
(sysv) follow after that 12 second delay (and yes, just a summary of
WHICH bootscript runs after the delay - I know what it is like trying
to write down the messages once the bootscripts are running.)
cat /var/log/boot.log
Is it created if /sbin/init is not called?

Pierre
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mai
Frans de Boer
2018-09-28 07:57:26 UTC
Permalink
Post by Pierre Labastie
Post by Bruce Dubbs
Post by Frans de Boer
Ken and Bruce,
I suspected it has to do something with entropy, so I used 4.9.88 - 4.9.2
dit not compiled - and the result was the same. So, since 4.9.88 is from
somewhere in March I assume that there is no provision yet for the spectre
etc. issue.
4.9 ?  Really ?  Rather you than me, I will be surprised if it works
with gcc-8.2.
Post by Frans de Boer
And yes, of course I did "modules_install", the modules are installed and
the initramfs is generated properly too. Until August all worked well, but
after August it did not anymore.
We seem to be at cross purposes.  You said you were using the
development book, now you mention an initramfs so clearly you have
at least something from BLFS.  So, specifying the version(s) of the
books might be relevant.  I've only booted (sysv) later than 8.3 on
two systems, one with 4.14.71, the other with 4.18.9, using
LFS-20180920 and BLFS-20180915 with occasional newer versions,
and neither have the entropy problem.
BUT: with 4.18.x and your delay of 12 seconds - What boot messages
(sysv) follow after that 12 second delay (and yes, just a summary of
WHICH bootscript runs after the delay - I know what it is like trying
to write down the messages once the bootscripts are running.)
cat /var/log/boot.log
Is it created if /sbin/init is not called?
Pierre
Nope, it is not.

Frans
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

http://en.wiki
Bruce Dubbs
2018-09-28 15:50:51 UTC
Permalink
Post by Pierre Labastie
Post by Bruce Dubbs
Post by Frans de Boer
Ken and Bruce,
I suspected it has to do something with entropy, so I used 4.9.88 - 4.9.2
dit not compiled - and the result was the same. So, since 4.9.88 is from
somewhere in March I assume that there is no provision yet for the spectre
etc. issue.
4.9 ?  Really ?  Rather you than me, I will be surprised if it works
with gcc-8.2.
Post by Frans de Boer
And yes, of course I did "modules_install", the modules are installed and
the initramfs is generated properly too. Until August all worked well, but
after August it did not anymore.
We seem to be at cross purposes.  You said you were using the
development book, now you mention an initramfs so clearly you have
at least something from BLFS.  So, specifying the version(s) of the
books might be relevant.  I've only booted (sysv) later than 8.3 on
two systems, one with 4.14.71, the other with 4.18.9, using
LFS-20180920 and BLFS-20180915 with occasional newer versions,
and neither have the entropy problem.
BUT: with 4.18.x and your delay of 12 seconds - What boot messages
(sysv) follow after that 12 second delay (and yes, just a summary of
WHICH bootscript runs after the delay - I know what it is like trying
to write down the messages once the bootscripts are running.)
cat /var/log/boot.log
Is it created if /sbin/init is not called?
No, it is not, but the discussion above asks about 'WHICH bootscript'
and I was replying to that.

I agree that the original problem was kernel configuration that caused a
failure before /sbin/init is called.

-- Bruce
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

http://en.wikipedia.org/wiki/Postin
Frans de Boer
2018-09-29 07:00:53 UTC
Permalink
Post by Bruce Dubbs
Post by Pierre Labastie
Post by Bruce Dubbs
Post by Frans de Boer
Ken and Bruce,
I suspected it has to do something with entropy, so I used 4.9.88 - 4.9.2
dit not compiled - and the result was the same. So, since 4.9.88 is from
somewhere in March I assume that there is no provision yet for the spectre
etc. issue.
4.9 ?  Really ?  Rather you than me, I will be surprised if it works
with gcc-8.2.
Post by Frans de Boer
And yes, of course I did "modules_install", the modules are installed and
the initramfs is generated properly too. Until August all worked well, but
after August it did not anymore.
We seem to be at cross purposes.  You said you were using the
development book, now you mention an initramfs so clearly you have
at least something from BLFS.  So, specifying the version(s) of the
books might be relevant.  I've only booted (sysv) later than 8.3 on
two systems, one with 4.14.71, the other with 4.18.9, using
LFS-20180920 and BLFS-20180915 with occasional newer versions,
and neither have the entropy problem.
BUT: with 4.18.x and your delay of 12 seconds - What boot messages
(sysv) follow after that 12 second delay (and yes, just a summary of
WHICH bootscript runs after the delay - I know what it is like trying
to write down the messages once the bootscripts are running.)
cat /var/log/boot.log
Is it created if /sbin/init is not called?
No, it is not, but the discussion above asks about 'WHICH bootscript'
and I was replying to that.
I agree that the original problem was kernel configuration that caused
a failure before /sbin/init is called.
  -- Bruce
Before sleeping, it struck me: it's QEMU version 3.0 causing this
problem. It seems that this version is allergic to the '-cpu host'
argument. Once I removed that, it worked fine!. I had changed this in
most scripts, but not yet the script to start LFS/BLFS.

I apologize to all to bother you with a problem I knew how to solve
beforehand.

Until the next /real/ issue,
Frans.

Paul Rogers
2018-09-28 22:38:21 UTC
Permalink
Re: the entropy issue - It's always a good idea to "dd" the host's random number seed to a new system to start off with. Storage locations may vary.
--
Paul Rogers
***@fastmail.fm
Rogers' Second Law: "Everything you do communicates."
(I do not personally endorse any additions after this line. TANSTAAFL :-)
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

http:/
Continue reading on narkive:
Loading...